Download Full Issue in PDF - Academy Publisher
Download Full Issue in PDF - Academy Publisher
Download Full Issue in PDF - Academy Publisher
Create successful ePaper yourself
Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.
1408 JOURNAL OF COMPUTERS, VOL. 8, NO. 6, JUNE 2013<br />
new solution. State entry 9, 2, 3, and 7 from the best<br />
solution form part of new solution. State entry 8, 3, 9, 2, 4<br />
and 6 from current solution form the other part of new<br />
solution. Modification of the new solution is carried out<br />
to make sure that duplicated state entry 2, 3, and 9 do not<br />
appear <strong>in</strong> the solution. By do<strong>in</strong>g so, unassigned state<br />
entry 0, 1 and 5 are filled <strong>in</strong>.<br />
Figure 2. Exchange method.<br />
Figure 3. Swap method for two states close to each other.<br />
Figure 4. Swap method for two states apart from each other.<br />
D. Crossover<br />
Crossover operator is designed to escape the local<br />
optimum <strong>in</strong> the GA. In this paper crossover operator is<br />
adapted <strong>in</strong> the simulated anneal<strong>in</strong>g to generate a new<br />
solution by comb<strong>in</strong><strong>in</strong>g the best solution with the current<br />
solution. Consequently, some potential parts of the<br />
solution can be <strong>in</strong>herited from the previous solution.<br />
It operates as a position-based crossover [18]. An array<br />
of b<strong>in</strong>ary bits is <strong>in</strong>itialised randomly, <strong>in</strong> which the length<br />
of the array equals to the length of the representation.<br />
When 1s appear <strong>in</strong> the array, copy the states from the best<br />
solution to new solution. When 0s appear <strong>in</strong> the array,<br />
copy the states from the current solution to new solution.<br />
S<strong>in</strong>ce each entry of state assignment solution is unique,<br />
cont<strong>in</strong>uous check is required avoid<strong>in</strong>g the repetition of<br />
the same states which is not allowed. If state entry is<br />
duplicated, the first unassigned state entry is assigned.<br />
Figure 5 shows an example of relay operator to generate<br />
Figure 5. Crossover operator.<br />
E. Cost Function<br />
By m<strong>in</strong>imis<strong>in</strong>g (6), low power dissipation could be<br />
achieved. Unfortunately, this may lead to area overhead,<br />
result<strong>in</strong>g <strong>in</strong> power overhead <strong>in</strong> the comb<strong>in</strong>ational part of<br />
the circuit. Therefore, objectives of area and power<br />
should be optimised simultaneously. The total cost<br />
function is C total = γC area + (1-γ) E, where C area is area<br />
function, E is power function and γ is a parameter<br />
specify<strong>in</strong>g the tradeoff of E with respect to C area .<br />
F. Proposed Algorithm<br />
To beg<strong>in</strong> with, an <strong>in</strong>itiation solution is randomly<br />
produced. The anneal<strong>in</strong>g approach is divided <strong>in</strong>to rough<br />
anneal<strong>in</strong>g and focus<strong>in</strong>g anneal<strong>in</strong>g, which is identified by<br />
Temperature threshold T threshold . In the rough anneal<strong>in</strong>g<br />
stage the temperature is high and uses exchange method.<br />
In the focus<strong>in</strong>g anneal<strong>in</strong>g stage the temperature is low<br />
and uses swap methods. The solution is evaluated by the<br />
cost function. The solution is accepted accord<strong>in</strong>g to a<br />
probability P = exp (-∆C/T), where ∆C is the difference<br />
cost between new solution and current solution, T is the<br />
temperature. P is between 0 and 1. If the solution is<br />
improved and kept accord<strong>in</strong>g to P, the new solution will<br />
be compared with the best recorded solution to decide<br />
whether updat<strong>in</strong>g the best record is necessary. If new<br />
solution is rejected <strong>in</strong>stead, this solution is then ignored<br />
and reverses the current solution. Outer loop updates the<br />
temperature. The temperature schedules at each outer<br />
iteration. In the <strong>in</strong>ner loop it generates possible solutions<br />
at each temperature. The algorithm term<strong>in</strong>ates when<br />
reach<strong>in</strong>g the lowest temperature and the output will be the<br />
best record. The pseudo code is shown <strong>in</strong> Algorithm 1.<br />
© 2013 ACADEMY PUBLISHER